The Annotation Process in the Turkish Treebank
I did not write a summary: the paper text that was sent has no content. It contains only the page markers <!-- с. 1 --> … <!-- с. 6 -->, with nothing between them. It looks like no text was extracted from the PDF. It may be a scan without a text layer, or the extractor may have failed on the fonts.
A summary written from this input would have to take its numbers and details from memory, with no way to check them against the text. For the Mova library, invented facts like these are worse than a missing summary.
To proceed, you can:
- Send the text again: extract it with a different tool, or run the PDF through OCR if it is a scan.
- Accept a draft based on general knowledge of the METU-Sabancı Turkish Treebank. Its sentence count, the dependencies between inflectional groups and the annotation stages would then need to be checked against the original, and the library entry should note that the summary is unverified.
Once the actual text is available, I will write the summary right away in the "## Summary" / "## Key points" format.
Cited by
- Coordination Structures in Dependency Treebanks
- Validation Issues induced by an Automatic Pre-Annotation Mechanism in the Building of Non-projective Dependency Treebanks
- Overcoming the challenges in morphological annotation of Turkish in universal dependencies framework
- Finding Function in Form: Compositional Character Models for Open Vocabulary Word Representation
- Labeled Pseudo-Projective Dependency Parsing with Support Vector Machines
Cite
@inproceedings{atalay2003annotation,
title = {{The Annotation Process in the Turkish Treebank}},
author = {Atalay, Nart B. and Oflazer, Kemal and Say, Bilge},
booktitle = {Proceedings of 4th International Workshop on Linguistically Interpreted Corpora (LINC-03) at EACL 2003},
year = {2003},
url = {https://aclanthology.org/W03-2405/}
}This page in Markdown → https://md.research.elamur.ai/free/papers/atalay-2003-annotation-process-turkish-treebank/